Image Restoration Algorithm For Space-variant Blur

The image is the most commonly used information expression vector in modern human social life.However,the digital image is often affected by external noise in the process of generation,transmission and storage.it would result in blurring even distortion of images.High-quality images are needed in many aspects such as computer vision,intelligent transportation,and medicine.Research on image deblurring algorithms is imperative.Since the 1960 s,with the deepening of research,people's understanding of fuzzy images has been continuously improved,and image deblurring algorithms have gradually matured to achieve better results.However,most of the image deblurring research in recent years are based on space-invariant model,algorithms based on the space-variant model is relatively few.Therefore,this topic is challenging and has corresponding academic significance and application value.In terms of academics,it is conducive to promoting the development of image processing and related disciplines.In terms of applications,it satisfies people's requirements for image clarity,and has played an active role in military and civilian fields.The image deblurring algorithm based on regularization has been quite mature.This paper starts with the formation of fuzzy images,and mainly studies the spatially-variant image deblurring algorithm.The main work and research results are as follows:1.This paper summarized mainstream image deblurring algorithm.Several typical image deblurring algorithms based on regularization are studied deeply.Analyzing the problems in the algorithm and the technical problems to be solved.2.Aiming at solving the loss of detail information of the flat area caused by the regular TV regularization algorithm,and the staircasing effect in the restored image.A general regularization model based on total variation and wavelet framework dual regularization is proposed.The ADMM algorithm is used to decompose the original problem into an easy solution to solve the sub problem,which makes the algorithm converge quickly,and completes the restoration and optimization of the image during the iterative process.3.The results of the new method are verified by the simulation experiment.The contrast experiment results show that the new algorithm has better deblurring effect than the traditional algorithms and the algorithm proposed by [41].